Terrain　music,　a　unique　way　to　convey　topographic　information,　has　made　progress　in　the　field　of　geomorphology　in　recent　years.　However,　research　has　been　limited　to　a　single　watershed　scale.　To　overcome　this　limitation　and　capture　terrain　characteristics　across　multiple　scales,　this　study　proposed　a　novel　geomorphological　analysis　method-Terrain　Music　Spectrum　(TMS).　By　introducing　statistical　quantitative　indicators　to　describe　the　characteristics　of　TMS,　this　research　conducted　systematic　investigations　into　geomorphological　quantification　and　recognition　applications.　Taking　the　Loess　Plateau　as　the　study　area　and　utilizing　Digital　Elevation　Model　(DEM)　data,　TMS　models　were　constructed　for　several　representative　sample　regions,　enabling　quantitative　characterization　of　watershed　geomorphological　features.　The　proposed　method　was　further　applied　to　landform　recognition　tasks.　The　results　showed　that　the　recognition　outcomes　based　on　the　TMS　method　were　consistent　with　those　based　on　traditional　terrain　indicators,　achieving　an　overall　accuracy　of　86.11%　and　a　Kappa　coefficient　of　0.83.　This　indicated　that　the　TMS　method　was　effective　in　distinguishing　different　landform　types.　Further　analysis　revealed　that　the　two　methods　exhibit　complementary　strengths　in　identifying　various　landform　categories.　Through　their　integration,　classification　performance　was　enhanced,　with　overall　accuracy　improving　to　88.89%　and　the　Kappa　coefficient　rising　to　0.87.　Feature　importance　analysis　demonstrated　that　TMS　indicators　played　a　significant　role　in　the　integrated　model.　These　findings　suggest　that　the　introduction　of　the　TMS　method　offers　a　new　perspective　for　geomorphological　research　and　facilitates　interdisciplinary　innovation　in　terrain　analysis.